Qmimic: Model-based monitoring and diagnosis

Qmimic is a fault and noise tolerant monitoring and diagnosis system which detects faults by comparing observations with predictions from an approximate plant model. Because faults are difficult to predict and model, Qmimic uses qualitative and semi-quantitative models of faulty and normal plants which capture broad classes of behaviors. Faults are diagnose by a hypothesize and test mechanism which suggests changes to the current model of the plant, generate a plant model with a presumed fault, and tests the new model against the current readings. The system is tested on a simulated reactor with two controllers.